Linqi Yu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Linqi Yu has a total of 25 co-authors. Here are the top co-authors by number of shared publications:
- Hee-Chang Lim
School of Mechanical Engineering, Pusan National University, Busan, Republic of Korea
13 shared publications - Mustafa Zhuhair Gheni Yousif
School of Mechanical Engineering, Pusan National University 1 , 2, Busandaehak-ro 63beon-gil, Geumjeong-gu, Busan 46241
11 shared publications - Meng Zhang
School of Mechanical Engineering, Pusan National University, 2, Busandaehak-ro 63beon-gil, Geumjeong-gu, Busan 46241, Republic of Korea
8 shared publications - Ricardo Vinuesa
University of Michigan
3 shared publications Dan Zhou
2 shared publications- Jung Sub Lee
Pusan National University Hospital
2 shared publications
Latest Publications
- Open Access
- 25 September 2026
- Published by Elsevier in SSRN Electronic Journal
A conditional latent-space deep learning surrogate for spatiotemporal temperature and smoke-field prediction in compartmented building fires
Rapid prediction of temperature and smoke evolution in compartmented buildings is important for fire safety assessment, scenario screening, and emergency decision support, but repeated high-fidelity computational fluid dynamics simulations are computationally expensive when many ignition scenarios must be evaluated. This study develops a conditional latent-space deep learning surrogate for full-field spatiotemporal prediction of compartmented building fires. The artificial-intelligence contribution is a two-stage surrogate framework that combines multi-scale three-dimensional autoencoding, sinusoidal coordinate--time embedding, and a U-shaped convolutional latent predictor to map ignition location and fire-development time to compact fire-field representations. The engineering application is rapid prediction of three-dimensional temperature and relative smoke-concentration fields in a six-compartment building configuration derived from a full-scale container-based fire training facility. Fire Dynamics Simulator simulations are conducted to construct 108 training ignition scenarios and 30 additional unseen testing scenarios for interpolation and extrapolation evaluation. The learned latent states are decoded to reconstruct full three-dimensional temperature and smoke fields. Results show that the proposed surrogate reproduces the dominant thermal-growth, smoke-layer formation, and inter-compartment propagation patterns. Autoencoder reconstruction, ablation analysis, hazard-region metrics, temporal--spatial error decomposition, and computational-efficiency analysis further demonstrate that the latent-space design preserves the principal fire-field structures, improves prediction accuracy, identifies safety-relevant hazardous regions, and provides orders-of-magnitude speed-up over direct simulation. The proposed framework provides an efficient artificial-intelligence tool for rapid building-fire scenario screening and preliminary hazard assessment.
- 25 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA conditional latent-space deep learning surrogate for spatiotemporal temperature and smoke-field prediction in compartmented building fires
Rapid prediction of temperature and smoke evolution in compartmented buildings is important for fire safety assessment, scenario screening, and emergency decision support, but repeated high-fidelity computational fluid dynamics simulations are computationally expensive when many ignition scenarios must be evaluated. This study develops a conditional latent-space deep learning surrogate for full-field spatiotemporal prediction of compartmented building fires. The artificial-intelligence contribution is a two-stage surrogate framework that combines multi-scale three-dimensional autoencoding, sinusoidal coordinate--time embedding, and a U-shaped convolutional latent predictor to map ignition location and fire-development time to compact fire-field representations. The engineering application is rapid prediction of three-dimensional temperature and relative smoke-concentration fields in a six-compartment building configuration derived from a full-scale container-based fire training facility. Fire Dynamics Simulator simulations are conducted to construct 108 training ignition scenarios and 30 additional unseen testing scenarios for interpolation and extrapolation evaluation. The learned latent states are decoded to reconstruct full three-dimensional temperature and smoke fields. Results show that the proposed surrogate reproduces the dominant thermal-growth, smoke-layer formation, and inter-compartment propagation patterns. Autoencoder reconstruction, ablation analysis, hazard-region metrics, temporal--spatial error decomposition, and computational-efficiency analysis further demonstrate that the latent-space design preserves the principal fire-field structures, improves prediction accuracy, identifies safety-relevant hazardous regions, and provides orders-of-magnitude speed-up over direct simulation. The proposed framework provides an efficient artificial-intelligence tool for rapid building-fire scenario screening and preliminary hazard assessment.
- 8 November 2025
- Published by Elsevier in Computers & Fluids
- Vol. 305
Three-dimensional super-resolution reconstruction of turbulent flow using 3D-ESRGAN with random sampling strategy
- 8 November 2025
- Published by Elsevier in Computers & Fluids
- Vol. 305
Three-dimensional super-resolution reconstruction of turbulent flow using 3D-ESRGAN with random sampling strategy
- Open Access
- 8 November 2025
- Published by Elsevier in Results in Engineering
- Vol. 28
Deep learning approach-based prediction of transfer functions in structure-acoustic coupling models
- 8 November 2025
- Published by Elsevier in Results in Engineering
- Vol. 28
Open AccessDeep learning approach-based prediction of transfer functions in structure-acoustic coupling models
- Open Access
- 1 December 2024
- Published by AIP Publishing in Physics of Fluids
- Vol. 36 (12)
Flow field recovery in restricted domains using a generative adversarial network framework
This study aims to reconstruct the complete flow field from spatially restricted domain data by utilizing an enhanced super-resolution generative adversarial network (ESRGAN) model. The difficulty in flow field reconstruction lies in accurately capturing and reconstructing large amounts of data under nonlinear, multi-scale, and complex flow while ensuring physical consistency and high computational efficiency. The ESRGAN model has a strong information mapping capability, capturing fluctuating features from local flow fields. The effectiveness of the model in reconstructing the whole domain flow field is validated by comparing instantaneous velocity fields, flow statistical properties, and probability density distributions. Using laminar bluff body flow from direct numerical simulation (DNS) as a priori case, the model successfully reconstructs the complete flow field from three non-overlapping limited regions, with flow statistical properties perfectly matching the original data. Validation of the power spectrum density for the reconstruction results also proves that the model could conform to the temporal behavior of the real complete flow field. Additionally, tests using DNS turbulent channel flow with a friction Reynolds number (Reτ=180) demonstrate the ability of the model to reconstruct turbulent fields, though the quality of results depends on the number of flow features in the local regions. Finally, the model is applied to reconstruct turbulence flow fields from particle image velocimetry (PIV) experimental measurements, using limited data from the near-wake region to reconstruct a larger field of view. The turbulence statistics closely match the experimental data, indicating that the model can serve as a reliable data-driven method to overcome PIV field-of-view limitations while saving computational costs.
- 1 December 2024
- Published by AIP Publishing in Physics of Fluids
- Vol. 36 (12)
Open AccessFlow field recovery in restricted domains using a generative adversarial network framework
This study aims to reconstruct the complete flow field from spatially restricted domain data by utilizing an enhanced super-resolution generative adversarial network (ESRGAN) model. The difficulty in flow field reconstruction lies in accurately capturing and reconstructing large amounts of data under nonlinear, multi-scale, and complex flow while ensuring physical consistency and high computational efficiency. The ESRGAN model has a strong information mapping capability, capturing fluctuating features from local flow fields. The effectiveness of the model in reconstructing the whole domain flow field is validated by comparing instantaneous velocity fields, flow statistical properties, and probability density distributions. Using laminar bluff body flow from direct numerical simulation (DNS) as a priori case, the model successfully reconstructs the complete flow field from three non-overlapping limited regions, with flow statistical properties perfectly matching the original data. Validation of the power spectrum density for the reconstruction results also proves that the model could conform to the temporal behavior of the real complete flow field. Additionally, tests using DNS turbulent channel flow with a friction Reynolds number (Reτ=180) demonstrate the ability of the model to reconstruct turbulent fields, though the quality of results depends on the number of flow features in the local regions. Finally, the model is applied to reconstruct turbulence flow fields from particle image velocimetry (PIV) experimental measurements, using limited data from the near-wake region to reconstruct a larger field of view. The turbulence statistics closely match the experimental data, indicating that the model can serve as a reliable data-driven method to overcome PIV field-of-view limitations while saving computational costs.
- Open Access
- 1 October 2024
- Published by AIP Publishing in Physics of Fluids
- Vol. 36 (10)
Self-supervised learning for effective denoising of flow fields
In this study, we proposed an efficient approach based on a deep learning (DL) denoising autoencoder (DAE) model for denoising noisy flow fields. The DAE operates on a self-learning principle and does not require clean data as training labels. Furthermore, investigations into the denoising mechanism of the DAE revealed that its bottleneck structure with a compact latent space enhances denoising efficacy. Meanwhile, we also developed a deep multiscale DAE for denoising turbulent flow fields. Furthermore, we used conventional noise filters to denoise the flow fields and performed a comparative analysis with the results from the DL method. The effectiveness of the proposed DL models was evaluated using direct numerical simulation data of laminar flow around a square cylinder and turbulent channel flow data at various Reynolds numbers. For every case, synthetic noise was augmented in the data. A separate experiment used particle-image velocimetry data of laminar flow around a square cylinder containing real noise to test DAE denoising performance. Instantaneous contours and flow statistical results were used to verify the alignment between the denoised data and ground truth. The findings confirmed that the proposed method could effectively denoise noisy flow data, including turbulent flow scenarios. Furthermore, the proposed method exhibited excellent generalization, efficiently denoising noise with various types and intensities.
- 1 October 2024
- Published by AIP Publishing in Physics of Fluids
- Vol. 36 (10)
Open AccessSelf-supervised learning for effective denoising of flow fields
In this study, we proposed an efficient approach based on a deep learning (DL) denoising autoencoder (DAE) model for denoising noisy flow fields. The DAE operates on a self-learning principle and does not require clean data as training labels. Furthermore, investigations into the denoising mechanism of the DAE revealed that its bottleneck structure with a compact latent space enhances denoising efficacy. Meanwhile, we also developed a deep multiscale DAE for denoising turbulent flow fields. Furthermore, we used conventional noise filters to denoise the flow fields and performed a comparative analysis with the results from the DL method. The effectiveness of the proposed DL models was evaluated using direct numerical simulation data of laminar flow around a square cylinder and turbulent channel flow data at various Reynolds numbers. For every case, synthetic noise was augmented in the data. A separate experiment used particle-image velocimetry data of laminar flow around a square cylinder containing real noise to test DAE denoising performance. Instantaneous contours and flow statistical results were used to verify the alignment between the denoised data and ground truth. The findings confirmed that the proposed method could effectively denoise noisy flow data, including turbulent flow scenarios. Furthermore, the proposed method exhibited excellent generalization, efficiently denoising noise with various types and intensities.
- 20 February 2024
- Published by American Physical Society (APS) in Physical Review Fluids
- Vol. 9 (2) , 024603
Predicting unavailable parameters from existing velocity fields of turbulent flows using a GAN-based model
In this study, an efficient deep-learning model is developed to predict unavailable parameters, e.g., streamwise velocity, temperature, and pressure from available velocity components. This model, termed mapping generative adversarial network (M-GAN), consists of a label information generator (LIG) and an enhanced super-resolution generative adversarial network. LIG can generate label information helping the model to predict different parameters. The GAN-based model receives the label information from LIG and existing velocity data to generate the unavailable parameters. Two-dimensional (2D) Rayleigh-Bénard flow and turbulent channel flow are used to evaluate the performance of M-GAN. First, M-GAN is trained and evaluated by two-dimensional direct numerical simulation (DNS) data of a Rayleigh-Bénard flow. From the results, it can be shown that M-GAN can predict temperature distribution from the two-dimensional velocities. Furthermore, DNS data of turbulent channel flow at two different friction Reynolds numbers and 550 are applied simultaneously to train the M-GAN and examine its predicting ability for the pressure fields and the streamwise velocity from the other two velocity components. The instantaneous and statistical results of the predicted data agree well with the DNS data, even for the flow at , indicating that M-GAN can be trained to learn the mapping function of the unknown fields with good interpolation capability.- 20 February 2024
- Published by American Physical Society (APS) in Physical Review Fluids
- Vol. 9 (2) , 024603
Predicting unavailable parameters from existing velocity fields of turbulent flows using a GAN-based model
In this study, an efficient deep-learning model is developed to predict unavailable parameters, e.g., streamwise velocity, temperature, and pressure from available velocity components. This model, termed mapping generative adversarial network (M-GAN), consists of a label information generator (LIG) and an enhanced super-resolution generative adversarial network. LIG can generate label information helping the model to predict different parameters. The GAN-based model receives the label information from LIG and existing velocity data to generate the unavailable parameters. Two-dimensional (2D) Rayleigh-Bénard flow and turbulent channel flow are used to evaluate the performance of M-GAN. First, M-GAN is trained and evaluated by two-dimensional direct numerical simulation (DNS) data of a Rayleigh-Bénard flow. From the results, it can be shown that M-GAN can predict temperature distribution from the two-dimensional velocities. Furthermore, DNS data of turbulent channel flow at two different friction Reynolds numbers and 550 are applied simultaneously to train the M-GAN and examine its predicting ability for the pressure fields and the streamwise velocity from the other two velocity components. The instantaneous and statistical results of the predicted data agree well with the DNS data, even for the flow at , indicating that M-GAN can be trained to learn the mapping function of the unknown fields with good interpolation capability.- Open Access
- 1 December 2023
- Published by AIP Publishing in Physics of Fluids
- Vol. 35 (12)
Optimizing flow control with deep reinforcement learning: Plasma actuator placement around a square cylinder
This study introduces a deep reinforcement learning-based flow control approach to enhance the efficiency of multiple plasma actuators on a square cylinder. The research seeks to adjust the control inputs of these actuators to diminish both drag and lift forces on the cylinder, ensuring flow stability in the process. The proposed model uses a two-dimensional direct numerical simulation of flow past a square cylinder to represent the environment. The control approach involves adjusting the AC voltage across three specific configurations of the plasma actuators. Initially tested at a Reynolds number (ReD) of 100, this strategy was later applied at ReD of 180. We observed a 97% reduction in the mean drag coefficient at ReD = 100 and a 99% reduction at ReD = 180. Furthermore, the findings suggest that increasing the Reynolds number makes it harder to mitigate vortex shedding using plasma actuators on just the cylinder's rear surface. However, an optimized configuration of these actuators can fully suppress vortex shedding under the proposed control scheme.
- 1 December 2023
- Published by AIP Publishing in Physics of Fluids
- Vol. 35 (12)
Open AccessOptimizing flow control with deep reinforcement learning: Plasma actuator placement around a square cylinder
This study introduces a deep reinforcement learning-based flow control approach to enhance the efficiency of multiple plasma actuators on a square cylinder. The research seeks to adjust the control inputs of these actuators to diminish both drag and lift forces on the cylinder, ensuring flow stability in the process. The proposed model uses a two-dimensional direct numerical simulation of flow past a square cylinder to represent the environment. The control approach involves adjusting the AC voltage across three specific configurations of the plasma actuators. Initially tested at a Reynolds number (ReD) of 100, this strategy was later applied at ReD of 180. We observed a 97% reduction in the mean drag coefficient at ReD = 100 and a 99% reduction at ReD = 180. Furthermore, the findings suggest that increasing the Reynolds number makes it harder to mitigate vortex shedding using plasma actuators on just the cylinder's rear surface. However, an optimized configuration of these actuators can fully suppress vortex shedding under the proposed control scheme.
- Open Access
- 13 October 2023
- Published by Cambridge University Press (CUP) in Journal of Fluid Mechanics
- Vol. 973, A12
Physics-constrained deep reinforcement learning for flow field denoising
A multi-agent deep reinforcement learning (DRL)-based model is presented in this study to reconstruct flow fields from noisy data. A combination of reinforcement learning with pixel-wise rewards, physical constraints represented by the momentum equation and the pressure Poisson equation, and the known boundary conditions is used to build a physics-constrained deep reinforcement learning (PCDRL) model that can be trained without the target training data. In the PCDRL model, each agent corresponds to a point in the flow field and learns an optimal strategy for choosing pre-defined actions. The proposed model is efficient considering the visualisation of the action map and the interpretation of the model operation. The performance of the model is tested by using direct numerical simulation-based synthetic noisy data and experimental data obtained by particle image velocimetry. Qualitative and quantitative results show that the model can reconstruct the flow fields and reproduce the statistics and the spectral content with commendable accuracy. Furthermore, the dominant coherent structures of the flow fields can be recovered by the flow fields obtained from the model when they are analysed using proper orthogonal decomposition and dynamic mode decomposition. This study demonstrates that the combination of DRL-based models and the known physics of the flow fields can potentially help solve complex flow reconstruction problems, which can result in a remarkable reduction in the experimental and computational costs.
- 13 October 2023
- Published by Cambridge University Press (CUP) in Journal of Fluid Mechanics
- Vol. 973, A12
Open AccessPhysics-constrained deep reinforcement learning for flow field denoising
A multi-agent deep reinforcement learning (DRL)-based model is presented in this study to reconstruct flow fields from noisy data. A combination of reinforcement learning with pixel-wise rewards, physical constraints represented by the momentum equation and the pressure Poisson equation, and the known boundary conditions is used to build a physics-constrained deep reinforcement learning (PCDRL) model that can be trained without the target training data. In the PCDRL model, each agent corresponds to a point in the flow field and learns an optimal strategy for choosing pre-defined actions. The proposed model is efficient considering the visualisation of the action map and the interpretation of the model operation. The performance of the model is tested by using direct numerical simulation-based synthetic noisy data and experimental data obtained by particle image velocimetry. Qualitative and quantitative results show that the model can reconstruct the flow fields and reproduce the statistics and the spectral content with commendable accuracy. Furthermore, the dominant coherent structures of the flow fields can be recovered by the flow fields obtained from the model when they are analysed using proper orthogonal decomposition and dynamic mode decomposition. This study demonstrates that the combination of DRL-based models and the known physics of the flow fields can potentially help solve complex flow reconstruction problems, which can result in a remarkable reduction in the experimental and computational costs.
- Open Access
- 1 August 2023
- Published by AIP Publishing in Physics of Fluids
- Vol. 35 (8)
A Swin-transformer-based model for efficient compression of turbulent flow data
This study proposes a novel deep-learning-based method for generating reduced representations of turbulent flows that ensures efficient storage and transfer while maintaining high accuracy during decompression. A Swin-transformer (ST) network combined with a physical constraints-based loss function is utilized to compress the turbulent flows with high compression ratios and then restore the data with underlying physical properties. The forced isotropic turbulence is used to demonstrate the ability of the ST-based model, where the instantaneous and statistical results show the excellent ability of the model to recover the flow data with a remarkable accuracy. Furthermore, the capability of the ST model is compared with a typical convolutional neural network-based auto-encoder (CNN-AE) by using the turbulent channel flow at two friction Reynolds numbers Reτ = 180 and 550. The results generated by the ST model are significantly more consistent with the direct numerical simulation data than those recovered by the CNN-AE, indicating the superior ability of the ST model to compress and restore the turbulent flow. This study also compares the compression performance of the ST model at different compression ratios (CR s) and finds that the model has low enough error even at very high CR. Additionally, the effect of transfer learning (TL) is investigated, showing that TL reduces the training time by 64% while maintaining high accuracy. The results illustrate for the first time that the Swin-transformer-based model incorporating a physically constrained loss function can compress and restore turbulent flows with the correct physics.
- 1 August 2023
- Published by AIP Publishing in Physics of Fluids
- Vol. 35 (8)
Open AccessA Swin-transformer-based model for efficient compression of turbulent flow data
This study proposes a novel deep-learning-based method for generating reduced representations of turbulent flows that ensures efficient storage and transfer while maintaining high accuracy during decompression. A Swin-transformer (ST) network combined with a physical constraints-based loss function is utilized to compress the turbulent flows with high compression ratios and then restore the data with underlying physical properties. The forced isotropic turbulence is used to demonstrate the ability of the ST-based model, where the instantaneous and statistical results show the excellent ability of the model to recover the flow data with a remarkable accuracy. Furthermore, the capability of the ST model is compared with a typical convolutional neural network-based auto-encoder (CNN-AE) by using the turbulent channel flow at two friction Reynolds numbers Reτ = 180 and 550. The results generated by the ST model are significantly more consistent with the direct numerical simulation data than those recovered by the CNN-AE, indicating the superior ability of the ST model to compress and restore the turbulent flow. This study also compares the compression performance of the ST model at different compression ratios (CR s) and finds that the model has low enough error even at very high CR. Additionally, the effect of transfer learning (TL) is investigated, showing that TL reduces the training time by 64% while maintaining high accuracy. The results illustrate for the first time that the Swin-transformer-based model incorporating a physically constrained loss function can compress and restore turbulent flows with the correct physics.
- Top Cited PapersOpen Access
- 14 February 2023
- Published by Cambridge University Press (CUP) in Journal of Fluid Mechanics
- Vol. 957, A6
A transformer-based synthetic-inflow generator for spatially developing turbulent boundary layers
This study proposes a newly developed deep-learning-based method to generate turbulent inflow conditions for spatially developing turbulent boundary layer (TBL) simulations. A combination of a transformer and a multiscale-enhanced super-resolution generative adversarial network is utilised to predict velocity fields of a spatially developing TBL at various planes normal to the streamwise direction. Datasets of direct numerical simulation (DNS) of flat plate flow spanning a momentum thickness-based Reynolds number, $Re_\theta = 661.5\unicode{x2013}1502.0$ , are used to train and test the model. The model shows a remarkable ability to predict the instantaneous velocity fields with detailed fluctuations and reproduce the turbulence statistics as well as spatial and temporal spectra with commendable accuracy as compared with the DNS results. The proposed model also exhibits a reasonable accuracy for predicting velocity fields at Reynolds numbers that are not used in the training process. With the aid of transfer learning, the computational cost of the proposed model is considered to be effectively low. Furthermore, applying the generated turbulent inflow conditions to an inflow–outflow simulation reveals a negligible development distance for the TBL to reach the target statistics. The results demonstrate for the first time that transformer-based models can be efficient in predicting the dynamics of turbulent flows. They also show that combining these models with generative adversarial networks-based models can be useful in tackling various turbulence-related problems, including the development of efficient synthetic-turbulent inflow generators.- 14 February 2023
- Published by Cambridge University Press (CUP) in Journal of Fluid Mechanics
- Vol. 957, A6
Top Cited PapersOpen AccessA transformer-based synthetic-inflow generator for spatially developing turbulent boundary layers
This study proposes a newly developed deep-learning-based method to generate turbulent inflow conditions for spatially developing turbulent boundary layer (TBL) simulations. A combination of a transformer and a multiscale-enhanced super-resolution generative adversarial network is utilised to predict velocity fields of a spatially developing TBL at various planes normal to the streamwise direction. Datasets of direct numerical simulation (DNS) of flat plate flow spanning a momentum thickness-based Reynolds number, $Re_\theta = 661.5\unicode{x2013}1502.0$ , are used to train and test the model. The model shows a remarkable ability to predict the instantaneous velocity fields with detailed fluctuations and reproduce the turbulence statistics as well as spatial and temporal spectra with commendable accuracy as compared with the DNS results. The proposed model also exhibits a reasonable accuracy for predicting velocity fields at Reynolds numbers that are not used in the training process. With the aid of transfer learning, the computational cost of the proposed model is considered to be effectively low. Furthermore, applying the generated turbulent inflow conditions to an inflow–outflow simulation reveals a negligible development distance for the TBL to reach the target statistics. The results demonstrate for the first time that transformer-based models can be efficient in predicting the dynamics of turbulent flows. They also show that combining these models with generative adversarial networks-based models can be useful in tackling various turbulence-related problems, including the development of efficient synthetic-turbulent inflow generators.